Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    INDEX
    Explanations

    words related to physics or engineering terminology

    oai_token-act-pair · gpt-3.5-turbo

    references to types of weight classes, particularly "lightweight" and "heavyweight."

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Negative Logits
    sis
    -0.77
    atche
    -0.76
    owitz
    -0.73
    leon
    -0.71
    itual
    -0.69
    itia
    -0.68
    GM
    -0.68
     Occupations
    -0.68
    thus
    -0.67
    rea
    -0.67
    POSITIVE LOGITS
     lightweight
    1.26
     heavyweight
    1.06
    weight
    0.97
    weights
    0.87
     minded
    0.86
     advoc
    0.82
     nodd
    0.82
    ailability
    0.81
     weights
    0.79
     unification
    0.78
    Act Density 0.006%

    No Known Activations